1 /*-
2 * SPDX-License-Identifier: BSD-3-Clause
3 *
4 * Copyright (c) 1992 Keith Muller.
5 * Copyright (c) 1992, 1993
6 * The Regents of the University of California. All rights reserved.
7 *
8 * This code is derived from software contributed to Berkeley by
9 * Keith Muller of the University of California, San Diego.
10 *
11 * Redistribution and use in source and binary forms, with or without
12 * modification, are permitted provided that the following conditions
13 * are met:
14 * 1. Redistributions of source code must retain the above copyright
15 * notice, this list of conditions and the following disclaimer.
16 * 2. Redistributions in binary form must reproduce the above copyright
17 * notice, this list of conditions and the following disclaimer in the
18 * documentation and/or other materials provided with the distribution.
19 * 3. Neither the name of the University nor the names of its contributors
20 * may be used to endorse or promote products derived from this software
21 * without specific prior written permission.
22 *
23 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33 * SUCH DAMAGE.
34 */
35
36 #include <sys/types.h>
37 #include <sys/stat.h>
38 #include <sys/time.h>
39 #include <sys/resource.h>
40 #include <err.h>
41 #include <errno.h>
42 #include <fcntl.h>
43 #include <locale.h>
44 #include <paths.h>
45 #include <signal.h>
46 #include <stdio.h>
47 #include <stdlib.h>
48 #include <string.h>
49 #include "pax.h"
50 #include "extern.h"
51 static int gen_init(void);
52
53 /*
54 * PAX main routines, general globals and some simple start up routines
55 */
56
57 /*
58 * Variables that can be accessed by any routine within pax
59 */
60 int act = DEFOP; /* read/write/append/copy */
61 FSUB *frmt = NULL; /* archive format type */
62 int cflag; /* match all EXCEPT pattern/file */
63 int cwdfd; /* starting cwd */
64 int dflag; /* directory member match only */
65 int iflag; /* interactive file/archive rename */
66 int kflag; /* do not overwrite existing files */
67 int lflag; /* use hard links when possible */
68 int nflag; /* select first archive member match */
69 int tflag; /* restore access time after read */
70 int uflag; /* ignore older modification time files */
71 int vflag; /* produce verbose output */
72 int Dflag; /* same as uflag except for inode change time */
73 int Hflag; /* follow command line symlinks (write only) */
74 int Lflag; /* follow symlinks when writing */
75 int Oflag; /* limit to single volume */
76 int Xflag; /* archive files with same device id only */
77 int Yflag; /* same as Dflg except after name mode */
78 int Zflag; /* same as uflg except after name mode */
79 int vfpart; /* is partial verbose output in progress */
80 int patime = 1; /* preserve file access time */
81 int pmtime = 1; /* preserve file modification times */
82 int nodirs; /* do not create directories as needed */
83 int pmode; /* preserve file mode bits */
84 int pids; /* preserve file uid/gid */
85 int rmleadslash = 0; /* remove leading '/' from pathnames */
86 int exit_val; /* exit value */
87 int docrc; /* check/create file crc */
88 char *dirptr; /* destination dir in a copy */
89 const char *argv0; /* root of argv[0] */
90 sigset_t s_mask; /* signal mask for cleanup critical sect */
91 FILE *listf; /* file pointer to print file list to */
92 char *tempfile; /* tempfile to use for mkstemp(3) */
93 char *tempbase; /* basename of tempfile to use for mkstemp(3) */
94
95 /*
96 * PAX - Portable Archive Interchange
97 *
98 * A utility to read, write, and write lists of the members of archive
99 * files and copy directory hierarchies. A variety of archive formats
100 * are supported (some are described in POSIX 1003.1 10.1):
101 *
102 * ustar - 10.1.1 extended tar interchange format
103 * cpio - 10.1.2 extended cpio interchange format
104 * tar - old BSD 4.3 tar format
105 * binary cpio - old cpio with binary header format
106 * sysVR4 cpio - with and without CRC
107 *
108 * This version is a superset of IEEE Std 1003.2b-d3
109 *
110 * Summary of Extensions to the IEEE Standard:
111 *
112 * 1 READ ENHANCEMENTS
113 * 1.1 Operations which read archives will continue to operate even when
114 * processing archives which may be damaged, truncated, or fail to meet
115 * format specs in several different ways. Damaged sections of archives
116 * are detected and avoided if possible. Attempts will be made to resync
117 * archive read operations even with badly damaged media.
118 * 1.2 Blocksize requirements are not strictly enforced on archive read.
119 * Tapes which have variable sized records can be read without errors.
120 * 1.3 The user can specify via the non-standard option flag -E if error
121 * resync operation should stop on a media error, try a specified number
122 * of times to correct, or try to correct forever.
123 * 1.4 Sparse files (lseek holes) stored on the archive (but stored with blocks
124 * of all zeros will be restored with holes appropriate for the target
125 * file system
126 * 1.5 The user is notified whenever something is found during archive
127 * read operations which violates spec (but the read will continue).
128 * 1.6 Multiple archive volumes can be read and may span over different
129 * archive devices
130 * 1.7 Rigidly restores all file attributes exactly as they are stored on the
131 * archive.
132 * 1.8 Modification change time ranges can be specified via multiple -T
133 * options. These allow a user to select files whose modification time
134 * lies within a specific time range.
135 * 1.9 Files can be selected based on owner (user name or uid) via one or more
136 * -U options.
137 * 1.10 Files can be selected based on group (group name or gid) via one o
138 * more -G options.
139 * 1.11 File modification time can be checked against existing file after
140 * name modification (-Z)
141 *
142 * 2 WRITE ENHANCEMENTS
143 * 2.1 Write operation will stop instead of allowing a user to create a flawed
144 * flawed archive (due to any problem).
145 * 2.2 Archives written by pax are forced to strictly conform to both the
146 * archive and pax the specific format specifications.
147 * 2.3 Blocking size and format is rigidly enforced on writes.
148 * 2.4 Formats which may exhibit header overflow problems (they have fields
149 * too small for large file systems, such as inode number storage), use
150 * routines designed to repair this problem. These techniques still
151 * conform to both pax and format specifications, but no longer truncate
152 * these fields. This removes any restrictions on using these archive
153 * formats on large file systems.
154 * 2.5 Multiple archive volumes can be written and may span over different
155 * archive devices
156 * 2.6 A archive volume record limit allows the user to specify the number
157 * of bytes stored on an archive volume. When reached the user is
158 * prompted for the next archive volume. This is specified with the
159 * non-standard -B flag. The limit is rounded up to the next blocksize.
160 * 2.7 All archive padding during write use zero filled sections. This makes
161 * it much easier to pull data out of flawed archive during read
162 * operations.
163 * 2.8 Access time reset with the -t applies to all file nodes (including
164 * directories).
165 * 2.9 Symbolic links can be followed with -L (optional in the spec).
166 * 2.10 Modification or inode change time ranges can be specified via
167 * multiple -T options. These allow a user to select files whose
168 * modification or inode change time lies within a specific time range.
169 * 2.11 Files can be selected based on owner (user name or uid) via one or more
170 * -U options.
171 * 2.12 Files can be selected based on group (group name or gid) via one o
172 * more -G options.
173 * 2.13 Symlinks which appear on the command line can be followed (without
174 * following other symlinks; -H flag)
175 *
176 * 3 COPY ENHANCEMENTS
177 * 3.1 Sparse files (lseek holes) can be copied without expanding the holes
178 * into zero filled blocks. The file copy is created with holes which are
179 * appropriate for the target file system
180 * 3.2 Access time as well as modification time on copied file trees can be
181 * preserved with the appropriate -p options.
182 * 3.3 Access time reset with the -t applies to all file nodes (including
183 * directories).
184 * 3.4 Symbolic links can be followed with -L (optional in the spec).
185 * 3.5 Modification or inode change time ranges can be specified via
186 * multiple -T options. These allow a user to select files whose
187 * modification or inode change time lies within a specific time range.
188 * 3.6 Files can be selected based on owner (user name or uid) via one or more
189 * -U options.
190 * 3.7 Files can be selected based on group (group name or gid) via one o
191 * more -G options.
192 * 3.8 Symlinks which appear on the command line can be followed (without
193 * following other symlinks; -H flag)
194 * 3.9 File inode change time can be checked against existing file before
195 * name modification (-D)
196 * 3.10 File inode change time can be checked against existing file after
197 * name modification (-Y)
198 * 3.11 File modification time can be checked against existing file after
199 * name modification (-Z)
200 *
201 * 4 GENERAL ENHANCEMENTS
202 * 4.1 Internal structure is designed to isolate format dependent and
203 * independent functions. Formats are selected via a format driver table.
204 * This encourages the addition of new archive formats by only having to
205 * write those routines which id, read and write the archive header.
206 */
207
208 /*
209 * main()
210 * parse options, set up and operate as specified by the user.
211 * any operational flaw will set exit_val to non-zero
212 * Return: 0 if ok, 1 otherwise
213 */
214
215 int
main(int argc,char * argv[])216 main(int argc, char *argv[])
217 {
218 const char *tmpdir;
219 size_t tdlen;
220
221 (void) setlocale(LC_ALL, "");
222 listf = stderr;
223 /*
224 * Keep a reference to cwd, so we can always come back home.
225 */
226 cwdfd = open(".", O_RDONLY | O_CLOEXEC);
227 if (cwdfd < 0) {
228 syswarn(0, errno, "Can't open current working directory.");
229 return(exit_val);
230 }
231
232 /*
233 * Where should we put temporary files?
234 */
235 if ((tmpdir = getenv("TMPDIR")) == NULL || *tmpdir == '\0')
236 tmpdir = _PATH_TMP;
237 tdlen = strlen(tmpdir);
238 while (tdlen > 0 && tmpdir[tdlen - 1] == '/')
239 tdlen--;
240 tempfile = malloc(tdlen + 1 + sizeof(_TFILE_BASE));
241 if (tempfile == NULL) {
242 paxwarn(1, "Cannot allocate memory for temp file name.");
243 return(exit_val);
244 }
245 if (tdlen)
246 memcpy(tempfile, tmpdir, tdlen);
247 tempbase = tempfile + tdlen;
248 *tempbase++ = '/';
249
250 /*
251 * parse options, determine operational mode, general init
252 */
253 options(argc, argv);
254 if ((gen_init() < 0) || (tty_init() < 0))
255 return(exit_val);
256
257 /*
258 * select a primary operation mode
259 */
260 switch (act) {
261 case EXTRACT:
262 extract();
263 break;
264 case ARCHIVE:
265 archive();
266 break;
267 case APPND:
268 if (gzip_program != NULL)
269 err(1, "can not gzip while appending");
270 append();
271 break;
272 case COPY:
273 copy();
274 break;
275 default:
276 case LIST:
277 list();
278 break;
279 }
280 return(exit_val);
281 }
282
283 /*
284 * sig_cleanup()
285 * when interrupted we try to do whatever delayed processing we can.
286 * This is not critical, but we really ought to limit our damage when we
287 * are aborted by the user.
288 * Return:
289 * never....
290 */
291
292 void
sig_cleanup(int which_sig)293 sig_cleanup(int which_sig)
294 {
295 /*
296 * restore modes and times for any dirs we may have created
297 * or any dirs we may have read. Set vflag and vfpart so the user
298 * will clearly see the message on a line by itself.
299 */
300 vflag = vfpart = 1;
301 if (which_sig == SIGXCPU)
302 paxwarn(0, "Cpu time limit reached, cleaning up.");
303 else
304 paxwarn(0, "Signal caught, cleaning up.");
305
306 ar_close();
307 proc_dir();
308 if (tflag)
309 atdir_end();
310 exit(1);
311 }
312
313 /*
314 * setup_sig()
315 * set a signal to be caught, but only if it isn't being ignored already
316 */
317
318 static int
setup_sig(int sig,const struct sigaction * n_hand)319 setup_sig(int sig, const struct sigaction *n_hand)
320 {
321 struct sigaction o_hand;
322
323 if (sigaction(sig, NULL, &o_hand) < 0)
324 return (-1);
325
326 if (o_hand.sa_handler == SIG_IGN)
327 return (0);
328
329 return (sigaction(sig, n_hand, NULL));
330 }
331
332 /*
333 * gen_init()
334 * general setup routines. Not all are required, but they really help
335 * when dealing with a medium to large sized archives.
336 */
337
338 static int
gen_init(void)339 gen_init(void)
340 {
341 struct rlimit reslimit;
342 struct sigaction n_hand;
343
344 /*
345 * Really needed to handle large archives. We can run out of memory for
346 * internal tables really fast when we have a whole lot of files...
347 */
348 if (getrlimit(RLIMIT_DATA , &reslimit) == 0){
349 reslimit.rlim_cur = reslimit.rlim_max;
350 (void)setrlimit(RLIMIT_DATA , &reslimit);
351 }
352
353 /*
354 * should file size limits be waived? if the os limits us, this is
355 * needed if we want to write a large archive
356 */
357 if (getrlimit(RLIMIT_FSIZE , &reslimit) == 0){
358 reslimit.rlim_cur = reslimit.rlim_max;
359 (void)setrlimit(RLIMIT_FSIZE , &reslimit);
360 }
361
362 /*
363 * increase the size the stack can grow to
364 */
365 if (getrlimit(RLIMIT_STACK , &reslimit) == 0){
366 reslimit.rlim_cur = reslimit.rlim_max;
367 (void)setrlimit(RLIMIT_STACK , &reslimit);
368 }
369
370 /*
371 * not really needed, but doesn't hurt
372 */
373 if (getrlimit(RLIMIT_RSS , &reslimit) == 0){
374 reslimit.rlim_cur = reslimit.rlim_max;
375 (void)setrlimit(RLIMIT_RSS , &reslimit);
376 }
377
378 /*
379 * signal handling to reset stored directory times and modes. Since
380 * we deal with broken pipes via failed writes we ignore it. We also
381 * deal with any file size limit through failed writes. Cpu time
382 * limits are caught and a cleanup is forced.
383 */
384 if ((sigemptyset(&s_mask) < 0) || (sigaddset(&s_mask, SIGTERM) < 0) ||
385 (sigaddset(&s_mask,SIGINT) < 0)||(sigaddset(&s_mask,SIGHUP) < 0) ||
386 (sigaddset(&s_mask,SIGPIPE) < 0)||(sigaddset(&s_mask,SIGQUIT)<0) ||
387 (sigaddset(&s_mask,SIGXCPU) < 0)||(sigaddset(&s_mask,SIGXFSZ)<0)) {
388 paxwarn(1, "Unable to set up signal mask");
389 return(-1);
390 }
391 memset(&n_hand, 0, sizeof n_hand);
392 n_hand.sa_mask = s_mask;
393 n_hand.sa_flags = 0;
394 n_hand.sa_handler = sig_cleanup;
395
396 if (setup_sig(SIGHUP, &n_hand) ||
397 setup_sig(SIGTERM, &n_hand) ||
398 setup_sig(SIGINT, &n_hand) ||
399 setup_sig(SIGQUIT, &n_hand) ||
400 setup_sig(SIGXCPU, &n_hand))
401 goto out;
402
403 n_hand.sa_handler = SIG_IGN;
404 if ((sigaction(SIGPIPE, &n_hand, NULL) < 0) ||
405 (sigaction(SIGXFSZ, &n_hand, NULL) < 0))
406 goto out;
407 return(0);
408
409 out:
410 syswarn(1, errno, "Unable to set up signal handler");
411 return(-1);
412 }
413